G-protein-coupled receptors (GPCRs) are integral membrane proteins involved in important physiological processes, including cell-to-cell communication, mediation of hormonal activity and sensory transduction. Many GPCRs have been implicated as major therapeutic routes to the treatment of human diseases. Today a large number of pharmaceuticals, identified by screening compound libraries, target GPCRs. Despite the striking clinical relevance of GPCRs, only one high-resolution structure (bovine rhodopsin) is known, severely limiting rational drug design. A major factor contributing to the structure determination of rhodopsin was its availability in large quantities from natural sources. In contrast, recombinant overexpression followed by efficient purification methods are needed for GPCRs that occur naturally at low levels. The Protein Data Bank contains depositions for about 24,000 structures of soluble proteins, but only about 200 depositions for structures of integral inner and outer membrane proteins, almost solely from bacterial origin. To date, there is no structure available of a eukaryotic membrane protein, which was produced in a recombinant system ! We focus on the three-dimensional structure determination of (eukaryotic) GPCRs by x-ray crystallography. Integral membrane proteins such as GPCRs reside in hydrophobic lipid bilayers. Therefore, recombinant expression has to ensure that GPCRs are targeted and correctly inserted into the host membrane system. Purification of GPCRs must be carried out in the presence of detergents to keep receptors in solution. However, most GPCRs are unstable and denature once extracted from the membrane with detergents. Formation of diffracting crystals requires protein-protein contacts mediated by the hydrophilic regions of the receptor. However, the loops connecting the receptor transmembrane helices, essential for forming crystal contacts, are often small. To date, there is no structure available of a GPCR from a recombinant source. This reflects difficulty in (a) overexpressing receptors in functional form, (b) in maintaining functionality during the purification in the presence of detergents, (c) in developing large-scale purification procedures, and (d) in finding conditions that allow efficient formation of crystal contacts. GPCRs recognize a large number of distinct ligands, ranging from odorant molecules and neurotransmitters to hormones. Some small molecule ligands have been proposed to bind within the hydrophobic core of the respective receptor. In contrast, extracellular domains are involved in ligand recognition of e.g. peptide receptors. Currently, only the bovine rhodopsin structure in its ground state is known. Clearly, additional structures are needed to understand the specificity of ligand binding and G-protein coupling to receptors involved in particular diseases, and hence to obtain the tools to design effective chemotherapeutic agents. Most importantly, GPCR structures with and without ligand will show the different conformational receptor states, helping to elucidate the mechanism of signal transduction. Specific GPCR structures (other than that of rhodopsin) will have a high impact on health issues such as HIV-1 infection, diabetes and many other diseases. The development and optimization of heterologous expression and large-scale purification of GPCRs requires robust analysis tools to rigorously assess the quality (functionality) of expressed and purified receptors. In 1991, at the MRC Laboratory of Molecular Biology (Cambridge, England), we chose the rat high-affinity neurotensin receptor for these purposes because of the availability of a hydrophilic non-sticky radio-ligand ([3H]neurotensin) that allows us to determine the amount of functional receptors, not only in membrane-bound form, but also in the detergent-solubilized state with low background binding. Using neurotensin receptor as a model for methods development, we have established a bacterial expression system for the production of functional, membrane-inserted receptors as maltose-binding protein fusions, and developed a procedure for the purification of fully functional receptors in detergent solution. To show that the developed methods can also be used for other GPCRs, we have purified milligram quantities of an adenosine receptor using a similar approach. In collaboration with Joseph Shiloach (NIDDK Biotechnology Unit), we developed conditions for fermentation at the 200-liter scale to provide the material needed to feed into weekly purification routines. Next, we established a fully automated purification scheme for functional neurotensin receptor fusion proteins at the 3-milligram or 10-milligram level, using immobilized metal affinity chromatography and a neurotensin column. This provides high-quality receptor protein for crystallization experiments on a regular basis. In parallel, we pursued experiments to determine the conformation of the agonist neurotensin when bound to its receptor. This was accomplished in collaboration with Marc Baldus and Sorin Luca (Gottingen, Germany) by using solid-state NMR techniques. Functional receptors were purified in milligram amounts and subsequently reconstituted into lipid vesicles. Solid-state NMR experiments were optimized to allow the unequivocal detection of microgram quantities of [13C,15N]-labeled neurotensin(8-13) in complex with functional receptor. The NMR data are consistent with a disordered state of the ligand in the absence of receptor. Upon receptor binding, the peptide undergoes a linear rearrangement, adopting a beta-strand conformation. We have optimized expression constructs with tobacco etch virus protease recognition sites at either end of the receptor to allow the isolation of neurotensin receptor devoid of its fusion partners. These receptor preparations were used to generate monoclonal antibodies and we are exploring the effect of receptor?antibody fragment co-crystallization on crystal quality. Co-crystallization of antibody fragments with integral membrane proteins has successfully led to high-resolution 3D structures of a bacterial cytochrome oxidase, the yeast cytochrome bc1 complex, and bacterial ion channels. We believe that co-crystallization with antibody fragments may also be beneficial for the structure determination of GPCRs. Antibody fragments should not only improve the receptor stability in detergent solution and constrain receptor flexibility, but also mediate extended protein-protein contacts for crystal formation. Furthermore, the availability of high-affinity, conformation-specific antibody fragments will facilitate efficient purification protocols. We have used neurotensin receptor, devoid of its fusion partners, for immunization of mice with the goal of obtaining receptor-specific monoclonal antibodies. Hybridoma cell lines are now available, and antibodies will be characterized by Elisa and Western blot analyses, and by biochemical methods, to determine the binding specificity for neurotensin receptor.